This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Introduction
Chlorophyll-a (Chl-a) is the main pigment used by phytoplankton in photosynthesis to convert nutrients and carbon dioxide. The distribution of Chl-a in the ocean indicates the biophysical quality of water and the abundance of biological resources. Therefore, Chl-a has an important impact on marine fishery resources and their development in the coastal marine environment (Radiarta & Saitoh 2008). Chl-a distribution has been studied in various regions of the world ocean (Shetye et al. 1996; Sarangi et al. 2008; Singh & Chaturvedi 2010; Brewin et al. 2013; Zhang et al. 2017). However, it has not been studied in the shelf region of Bangladesh, which could be used to achieve the targets under the component (Goal 14: Life below water) of the Sustainable Development Goals (SDG) after the recent settlement of maritime border disputes.
A considerable understanding of temporal dynamics of biophysical factors describing coastal water systems is required to appropriately manage and monitor valuable coastal ecosystems. For this purpose, remote sensing could be an effective tool. High resolution, multispectral satellite data have been used in recent decades to detect regular chemical, biological, and physical phenomena in coastal waters. Nonetheless, monitoring of coastal waters using the existing satellite instruments is a challenging task and requires extremely sophisticated procedures (Chauhan et al. 2002).
Spatial ocean parameters, such as Chl-a, have been extensively studied using time series of satellite data derived from ocean color satellite measurements. The MODIS-Aqua satellite sensor is one of the most popular satellites providing data for the ocean. It is monitored by the National Aeronautics and Space Administration (NASA) and is popular among ocean dynamics researchers (Singh & Chaturvedi 2010).
Several efforts have been made to study Chl-a variability in the Arabian Sea, which is adjacent to the Bay of Bengal (BoB; Sathyendranath et al. 1991; Bhattathiri et al. 1996). Although some studies discuss the Chl-a distribution in the BoB, most of them are based on the southwestern and western parts of BoB (Prasad & Singh 2010; Nagamani et al. 2011; Nagamani et al. 2013; Poornima et al. 2013; Suwannathatsa & Wongwises 2013). Thus, no studies were found that would focus particularly on the shelf region of the northern part of the BoB. Therefore, we took the first step toward understanding the variability of Chl-a concentration in this region.
The shelf region of Bangladesh is relatively flat without being cut-off by a submarine canyon and extends for approximately 150–200 km with a depth of 150–200 m (Qasim 1977). The region is unique due to an enormous freshwater flux from surrounding rivers and seasonally reversed monsoon winds, and exerts a strong impact on the surface water circulation and stratification (Sarangi et al. 2008). During monsoons, large volumes of freshwater discharged by the Ganga-Brahmaputra-Meghna rivers significantly reduce the salinity and intensity of upwelling at a distance of up to ~40 km from the coast (Shetye et al. 1996). Therefore, the monsoon and river discharge may affect the Chl-a distribution in the shelf region. In general, the sea surface temperature (SST) has an inverse relation with Chl-a distribution and the trend in this relation is consistent mainly in the offshore region, but not always in the shelf region (Khalil et al. 2009). Thus, a baseline study is required.
This study, therefore, aimed to investigate annual and seasonal variability in Chl-a and its relationship with SST and river discharge in the shelf region of the northern BoB and validates the MODIS-Aqua satellite data against in-situ data.
Materials and methods
The study focused on the shelf region of the northern BoB (latitude: from 20°N to 23°N; longitude: from 88.77°E to 92.70°E), which extends up to a depth of 200 m (Fig. 1).
Figure 1
Shelf region of the northern BoB including 10 sampling locations
Chl-a, SST, and river discharge data were considered in this study. MODIS-Aqua Level 3 mapped 4 km resolution monthly and annual Chl-a concentration and SST data for 2002–2018 were retrieved from the Ocean Biology Distributed Active Archive Center (OB. DAAC) of NASA (https://oceancolor.gsfc.nasa.gov/l3/).
MODIS-Aqua Chl-a data use the OCI (Ocean Color Index) algorithm to derive Chl-a values from the recorded radiance. The OCI algorithm is a combination of two algorithms, the O'Reilly band ratio OCx (chl_ oc4) algorithm (O'Reilly et al. 1998) and the Hu color index (CI) algorithm (Hu et al. 2012).
The SST data are generated using both 11 and 12 μm long wave infrared bands. The current SST algorithm is based on a modified version of the nonlinear SST algorithm of Walton et al. (1998). It is applicable to the MODIS sensor for both daytime and nighttime observations. In this study, daytime 11 μm data were used to analyze the SST values. The MODIS Aqua monthly and annual-scale Chl-a and SST datasets were processed and analyzed using SeaDAS 7.4.
River discharge data (2002–2018) were obtained from the Bangladesh Water Development Board (BWDB). River discharge data from two stations (SW273 and SW93.5L), which cover the major discharge into the BoB, were used to determine the impact of inland discharge on Chl-a concentration.
For in-situ measurements, the United Nations Educational, Scientific and Cultural Organization (UNESCO) monograph (Jeffrey & Mantoura 1997) was followed. Aminot & Rey (2001) was also followed for adequate quality assessment and quality control practices. The in-situ Chl-a determination was based on 90% acetone extraction. Water samples were collected from ten sampling locations covering a latitude of 21°34’28.8”N to 20°50’37.0”N and a longitude of 89°28’E, with a 5’ (5 min) geographical coordinate interval (Fig. 1). Water sampling dates for this study were 12 and 13 November 2017, 2 December 2017 and 7 January 2018 during daytime. A nitrocellulose filter of 47 mm diameter and 0.45 μm pore size was used in the filtration process. The extract volume was maintained at 10 ml of 90% acetone (i.e. 10 ml + dead volume of the filter). The absorbance of sample extracts was measured at 750 nm and 665 nm in a Hitachi U-2910 spectrophotometer against 90% acetone blank before and after acidification with 2 drops of 1N HCl. The absorbance readings obtained at 665 nm and 750 nm before and after acidification and cell to cell differences were entered in the following monochromatic equation of Lorenzen (1967), which is recommended for Chl-a estimation in coastal and estuarine waters:
A = absorption coefficient of chl-a (11.4 μg cm ml−1)
L = cuvette light-path in centimeter (cm)
Ve = extraction volume in milliliter (ml)
Vf = filtered volume in liter (l)
E 665o and E 750o = absorbance of 665 nm and 750 nm, respectively, before acidification
E 665a and E 750a = absorbance of 665 nm and 750 nm, respectively, after acidification
R = maximum absorbance ratio of E665o/E665a in the absence of pheopigments (1.7)
K = R / (R − 1) = 2.43
Results and discussion
Annual trend in Chl-a concentration
The annual time series data (2002–2018) acquired from the MODIS-Aqua satellite revealed a complex variability in Chl-a concentration in the shelf region of the northern BoB. The maximum and minimum annual average Chl-a concentration in the study area was 2.94 mg m−3 in 2004 and 2.08 mg m−3 in 2015, respectively. The annual chronological Chl-a satellite images from 2002 to 2018 are shown in Figure 2. Considering the spatial variability, we found a higher and wider range of Chl-a concentration (0.18–15.93 mg m-3) in the study area compared with other studies that were performed in the western (0.1–0.9 mg m−3; Nagamani et al. 2011) and southwestern (0.1–2.0 mg m−3; Sarangi et al. 2008) region of the BoB. The high Chl-a concentration is caused by the river discharge along the coast (Dasgupta et al. 2009) with a depth of 200 m combined with strong winds (Dey & Singh 2003; Patti et al. 2008; Dasgupta et al. 2009; Singh & Chaturvedi 2010).
Figure 2
MODIS-Aqua satellite-derived annual composite Chl-a concentration images from 2002 to 2018
Annual variations are generally influenced by physical properties (temperature, salinity, and dissolved inorganic nutrient concentration) of the ocean. Different regions of the ocean generally show different patterns of Chl-a concentration, which are affected by diverse factors, such as tidal effects, inconsistent upwelling and long-term environmental changes (i.e. SST). Upwelling can be an important factor affecting long-term changes in Chl-a concentration. In our study, the annual Chl-a concentration in the shelf region of the BoB showed a declining trend between 2002 and 2018, which follows a simple linear regression with a rate of −0.0183 mg m−3 yr−1 (Fig. 3). Behrenfeld et al. (2006) found a declining trend in Chl-a concentration over nine years and attributed it to several climatic factors, particularly SST, and some local factors. A similar finding was reported by Brodie et al. (2007).
Figure 3
Annual trend in Chl-a concentration from 2002 to 2018
Seasonal trend in Chl-a concentration
Chl-a concentration was higher (2.21 ± 0.56 mg m−3) during the northeast monsoon (October– February), within the range of 0.20–4.85 mg m−3, and lower (1.81±1.14 mg m−3) during the pre-monsoon (March–May), within the range of 0.27–5.94 mg m−3 (Fig. 4). The seasonal pattern of Chl-a concentration in the shelf region of the BoB is illustrated with monthly climatological composites of the MODIS-Aqua images from July 2002 to June 2018 (Fig. 5). Figures 4 and 5 clearly show that Chl-a concentration gradually increased from the end of the southwest monsoon (September), was mostly high during the northeast monsoon (October–February), and relatively low in pre-monsoon (March–May).
Figure 4
Seasonal trend in Chl-a concentration (July 2002 to June 2018)
Figure 5
MODIS-Aqua-derived monthly climatological Chl-a concentration images from July 2002 to June 2018 (SWM – southwest monsoon, NEM – northeast monsoon, PM – pre-monsoon)
The relatively high concentration of Chl-a (2.21 ± 0.56 mg m−3) in the northeast season represents a highly productive season. The high Chl-a concentration in this season is associated with run-off from the main rivers, low SST (~26°C), and coastal plumes (Chaturvedi 2005; Sarangi et al. 2008; Nagamani et al. 2011). Of the northeast season months, October and November showed higher Chl-a concentration (2.41 mg m−3 and 2.49 mg m−3, respectively). A higher river discharge (7578.55 m3 s−1) was observed in this region in July–September, whereas high Chl-a concentration was observed in September–November, thus indicating a lag time. This time lag was mainly caused by the time required for the ocean to respond to the wind, forcing the nutrient-rich water to reach the surface for the phytoplankton cells to grow (Sarangi & Devi 2017). Suwannathatsa & Wongwises (2013) also found high Chl-a concentration (5–15 mg m−3) along the coasts of Bangladesh from late October to January due to strong currents and river run-off. Several cyclones occur in the BoB in October–December, which provide the surface with nutrient-rich subsurface water, form eddies and intensify blooms, and consequently play a role in the formation of high Chl-a concentration (Chauhan et al. 2002; Madhu et al. 2002; Sarangi et al. 2008). Sarangi & Devi (2017) reported the maximum number of larger eddies, observed particularly in October, and indicated the eddy pumping as a possible mechanism of vertical transfer of nutrients across the halocline to the oligotrophic or eutrophic zone in the BoB. A similar observation was made by Bhushan et al. (2018), Krishna (2013), and Venkateswrlu & Rao (2004) in connection with a cyclonic disturbance. Nagamani et al. (2011) also observed dense Chl-a patches (~1.0 mg m−3) after a cyclone in the BoB in October 1999. Thus, the BoB receives much attention in regard to Chl-a due to the formation of severe tropical cyclones.
The lowest Chl-a concentration was found in the shelf region of the northern BoB during the pre-monsoon season (March–May). It occurred due to a gradual increase in SST (detailed relationship between Chl-a concentration and SST are described in the subsequent section) and a decrease in wind speed (Nagamani et al. 2013). The compensation of the northeast and southwest monsoons during the transition period from winter to summer neutralizes the propagated strong currents (Suwannathatsa & Wongwises 2013), which explains the decrease in Chl-a concentration to the lowest level in the year.
Relationship between Chl-a concentration and SST
The seasonal trend in Chl-a concentration and SST is shown in Figure 6 and the statistical relations are shown in Table 1. The linear regression analysis showed a weak inverse relation between the annual Chl-a concentration and SST. The inverse relationship between the Chl-a concentration and SST was found only during the pre-monsoon, while a positive relation was found during the northeast and southwest monsoon.
Figure 6
Seasonal trend in Chl-a concentration and SST
Statistical analysis of Chl-a concentration and SST
Time Series
a
b
r
R2
n
Southwest (June–September)
−5.505
0.254
0.224
0.050
61
Northeast (October–February)
0.345
0.070
0.283
0.080
80
Pre-monsoon (March–May)
6.604
−0.168
−0.186
0.035
47
Annual data (2002–2018)
8.951
−0.237
−0.326
0.160
17
The positive relation (coefficient of regression, b = 0.254) during the southwest monsoon (June–September) contradicts the usual inverse relation between Chl-a concentration and SST. Rainfall and the associated freshwater intrusion through the adjacent rivers provide the onshore region of the BoB with nutrient-rich (nitrate, silicon, and phosphate) freshwater containing nitrate concentration patches in June–August (Sarangi & Devi 2017). This nitrate concentration (influx water nutrient) was associated more with Chl-a concentration than SST, as nitrate inherently enhances the ocean Chl-a (Sarangi & Devi 2017). Therefore, the freshwater influx masked the effect of SST on the Chl-a concentration variability in this season.
A positive relation (b = 0.070) was also observed during the northeast monsoon (October–February) between Chl-a concentration and SST. In this season, the shelf region of the BoB was exposed to a relatively smaller amount of the river discharge than the southwest monsoon (Sarangi et al. 2008; Nagamani et al. 2011; Sarangi & Devi 2017). Thus, a lower b-value was observed in this season compared to the southwest monsoon. During the pre-monsoon (March–May), an inverse relation (b = −0.168) was observed between Chl-a concentration and SST. In this season, the river discharge rate was insignificant due to lower rainfall. Therefore, the impact of the river discharge on Chl-a concentration was extremely low, indicating that the river discharge did not mask the effect of SST in this season, as opposed to the southwest and northeast monsoon.
An inverse (b = −0.237) relation (Table 1) was also found between the annual Chl-a concentration and SST (Fig. 7). This inverse relationship was relatively weaker than in most scientific researches (Smyth et al. 2001; Behrenfeld et al. 2006; Brodie et al. 2007), which allowed us to conclude that SST has a weak relation with Chl-a concentration in the shelf region of the northern BoB.
Figure 7
Annual relationship between Chl-a concentration and SST
Relationship between Chl-a concentration and river discharge
The average annual river discharge between 2002 and 2018 was 6923 ± 915.37 m3 s−1. The highest amount of discharge was 9154 m3 s−1 in 2002, whereas the lowest amount was 5723 m3 s−1 in 2009. The higher amount of the river discharge was observed during the southwest monsoon (June–September) and it ranged from 4425 to 13329 m3 s−1 with an average of 7579 m3 s−1.
The Chl-a concentration and river discharge showed a positive relation (b = 0.004, R2 = 0.25, n = 17; Fig. 8). This positive relation is mostly due to the Ganga delta, which is the largest delta in the world (Coleman 1969; Kuehl et al. 1997; Goodbred et al. 2003), carrying anthropogenic deposits from thousands of kilometers of riverside settlements (Goodbred & Kuehl 2000; Islam & Gnauck 2008). The coefficient of determination (R2 = 0.25, n = 17) between Chl-a concentration and river discharge obtained in this study was relatively higher than the coefficient of determination (R2 = 0.16, n = 17) between Chl-a concentration and SST (see Table 1), which implies that the river discharge has a greater impact on Chl-a concentration than SST. Scholars agree that the river discharge has a greater effect on Chl-a concentration than other factors, e.g. upwelling, weather pattern, wind direction, floods, chemical compositions of water, biological parameters, geographical location, climatic factors (Johnson et al. 1999; Barnett et al. 2001; Levitus et al. 2001; Behrenfeld et al. 2006; Brodie et al. 2007; Subramaniam et al. 2008). Therefore, it is evident that the river discharge is one of the main driving factors affecting the variability of Chl-a concentration in the study area.
Figure 8
Relationship between Chl-a concentration and river discharge
Validation of satellite measurements by in-situ measurements
The correlation coefficient between the in-situ values and MODIS-Aqua satellite-derived Chl-a concentration was 0.78 (n = 120) with the R2 value of 0.70 (Fig. 9). Nagamani et al. (2013) observed a high correlation coefficient of 0.842 between the in-situ and satellite data in the coastal water of the BoB. Haëntjens et al. (2017) also found an R2 value of approximately 0.67 for the southern ocean, while compared the in-situ Chl-a concentration data with MODIS Aqua satellite Chl-a concentration data using the OCI algorithm.
Figure 9
Correlation between in-situ and MODIS Aqua satellite
Conclusions
This study presents the seasonal and annual variability in Chl-a concentration and its relationship with SST and river discharge in the shelf region of the northern BoB using MODIS-Aqua satellite data. A decrease in Chl-a concentration was observed between 2002 and 2018. The highest Chl-a concentration was observed during the northeast monsoon season and the lowest in the pre-monsoon season. Chl-a concentration showed a weak inverse relation with SST, both annually and seasonally, especially in the pre-monsoon season. In this region, the river discharge mainly affected the Chl-a concentration variability during the southwest and northeast monsoon. Thus, monsoon has a significant effect on Chl-a concentration variability. In addition, various parameters (nitrate, phosphate, silicon, wind speed, cyclones, and eddies) also affect the Chl-a variability in coastal water, which was not covered by this study. Though we performed the in-situ measurements to validate the MODIS data, further extensive studies are needed to ensure the reliability of MODIS-Aqua data through a regional correction or inversion algorithm.
